loh/fix elpa index

See merge request !10

lfind() expects a proper comparison function where the order of the
arguments does not matter. Previously it was assumed that the first
argument is always a pointer to an unspecific "key" object and the
second argument a pointer to an array member to compare against.

However, the standard requires the two arguments to be interchangeable
pointers to array-member like objects.

Surfaced as musl-libc does in fact call it the other way around,
compared to glibc.

Fix problem in elpa2_print_kernels with GPU kernel

See merge request !9

The GPU kernel can only be set, if e%set("use_gpu",1) has been set
before and GPU usage was thus requested.

In the elpa2_print_kernels, this was not reflected, and the GPU kernels
were checked via can_set method without activating GPU usage
=> GPU kernels were always reported as non exisitent

by allreducing the measured time and using average over ranks insted

to avoid stalling between storing and loading parameters

checking that the loaded values equal to the stored values

and removing one extra check

the error handling has to be thought of properly in some of the uses
of check_elpa_get/set (e.g. those outside the elpa_setup function)

mpi_comm_parent is allways requred (it was not required before, but
actually the code internals expected it to be supplied, at least for
ELPA 2 calculation OR whenever GPU was used)